Multimodal Fake News Classification -- 2

Fake news has altered society in negative ways in politics and culture. It has adversely affected both online social network systems as well as offline communities and conversations. Using automatic machine learning classification models is an efficient way to combat the widespread dissemination of fake news. While research in the area of fake news detection is of high importance for society, most of the existing methods opt for uni-mode approaches, as analyzing the multimodal content is often more complicated and availability of a reasonably sized dataset poses extra challenge. For this problem, the goal for you is to explore some baseline methods discussed in [1], where both text and image mode component of a sample are represented using their respective mode- specific feature descriptor. The mode specific features are later concatenated directly to obtain a multimodal feature representative for each sample and then propose improvement. More specifi- cally: • Baseline: The baseline you will implement will be the methods described by the authors of the dataset. Please feel free to choose your preferred word embeddings to represent the text component. You may also utilize several concepts learned in the course (like dimensionality reduction etc.) to demonstrate the effects on the results. • Extension: Based on your understanding, you are expected to improve or come up with a different learning structure to better handle the problem.